Skip to contentSkip to conversation

json-c: upgrade 0.18 -> 0.19

PATCH Superseded Moritz Haase 3 participants 6 messages Jun 29 – Jun 30

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 json-c: upgrade 0.18 -> 0.19

Release notes are available at [0].

[0]: https://github.com/json-c/json-c/blob/json-c-0.19/ChangeLog

---
 .../json-c/{json-c_0.18.bb => json-c_0.19.bb}               | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
 rename meta/recipes-devtools/json-c/{json-c_0.18.bb => json-c_0.19.bb} (85%)
meta/recipes-devtools/json-c/json-c_0.19.bb renamed from meta/recipes-devtools/json-c/json-c_0.18.bb +3 -3
similarity index 85%
rename from meta/recipes-devtools/json-c/json-c_0.18.bb
rename to meta/recipes-devtools/json-c/json-c_0.19.bb
index c112aacf4bceb0c41dbe2e1643d2d7e0944bf250..f16c7f1d59455e9ef19e7dffd5494b25f5d90e8f 100644
@@ -9,17 +9,17 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=de54b60fbbc35123ba193fea8ee216f2"
 SRC_URI = "https://s3.amazonaws.com/json-c_releases/releases/${BP}.tar.gz \
            file://run-ptest \
            "
-SRC_URI[sha256sum] = "876ab046479166b869afc6896d288183bbc0e5843f141200c677b3e8dfb11724"
+SRC_URI[sha256sum] = "37ad0249902e301bd9052bf712e511fcc6acff4ecaad4b5900aad9ce564e26de"
 
 # NVD uses full tag name including date
-CVE_VERSION = "0.18-20240915"
+CVE_VERSION = "0.19-20260627"
 
 UPSTREAM_CHECK_URI = "https://github.com/${BPN}/${BPN}/tags"
 UPSTREAM_CHECK_REGEX = "json-c-(?P<pver>\d+(\.\d+)+)-\d+"
 
 RPROVIDES:${PN} = "libjson"
 
-# Apps aren't needed/packaged and their CMakeLists.txt is incompatible with CMake 4+.
+# Apps aren't needed/packaged.
 EXTRA_OECMAKE = "-DDISABLE_WERROR=ON \
                  -DBUILD_APPS=OFF \
 "
#1 lore raw
Quoted 6 lines · Moritz Haase via lists.openembedded.org

Release notes are available at [0].

[0]: https://github.com/json-c/json-c/blob/json-c-0.19/ChangeLog

---

Hi Moritz,

Thanks for your patch.

It looks like the upgrade is breaking the ptests:

Failed ptests:
{'json-c': ['test_json_parse_cli.test']}

https://autobuilder.yoctoproject.org/valkyrie/#/builders/61/builds/3927
https://autobuilder.yoctoproject.org/valkyrie/#/builders/73/builds/3941
https://autobuilder.yoctoproject.org/valkyrie/#/builders/110/builds/624

Ptest logs can be found here:

https://valkyrie.yocto.io/pub/non-release/20260630-46/testresults/qemuarm64-ptest/core-image-ptest-json-c/
https://valkyrie.yocto.io/pub/non-release/20260630-46/testresults/qemux86-64-ptest/core-image-ptest-json-c/
https://valkyrie.yocto.io/pub/non-release/20260630-46/testresults/qemux86-64-musl-ptest/core-image-ptest-json-c/

I can see at least this line in the logs;

FAIL: test_json_parse_cli.test

Can you have a look at the issue?

Thanks,
Mathieu

Signature · 3 lines
Mathieu Dubois-Briand, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#2 lore raw

Hi Mathieu,

Quoted 5 lines

It looks like the upgrade is breaking the ptests:
[...]
I can see at least this line in the logs;

FAIL: test_json_parse_cli.test

upon checking the autobilder links, I couldn't find the test log file (json-c_test.log) that the ptest-runner generates (see [0]). Is it available somewhere or would I need to re-run the tests locally?

From looking at the implementation of the test (which was added in 0.19), I'd guess it fails because it calls the 'json_parse' app (see [1]) - and we don't build the apps (see 'EXTRA_OECMAKE').

If that is indeed the case, should the test be disabled/skipped? What's the preferred way to do this?

Regards,

Moritz

[0]: https://git.openembedded.org/openembedded-core/tree/meta/recipes-devtools/json-c/json-c/run-ptest?id=68f7ccf9f4c9bc4cd89b8c519f5b3b6f28a2464e#n13
[1]: https://github.com/json-c/json-c/blob/d9d8cab82291637671f6c90cf06e578750a08718/tests/test_json_parse_cli.test#L18

#3 lore raw
Quoted 6 lines · Moritz Haase via lists.openembedded.org

From looking at the implementation of the test (which was added in 0.19), I'd guess it fails because it calls the 'json_parse' app (see [1]) - and we don't build the apps (see 'EXTRA_OECMAKE').

If that is indeed the case, should the test be disabled/skipped? What's the preferred way to do this?

It would be better to enable the apps, package them separately, and make ptest package RDEPEND on them.

Alex

#4 lore raw

Hi,

Quoted 2 lines

It would be better to enable the apps, package them separately, and make ptest package RDEPEND on them.

just saw that I originally disabled the build of the apps as part of the CMake 4 migration, since they were not packaged and upstream doesn't consider them ready to be shipped (see [0]). This hasn't changed, the same text is still there and there are no install() rules for them.

Should we still try to package them? Even if we did that, the test itself would likely also require patching, as it tries to execute the json_parse app using a relative path (out of the build dir I guess).

Moritz

[0]: https://github.com/json-c/json-c/blob/7cee5237dc6c0831e3f9dc490394eaea44636861/apps/CMakeLists.txt#L119-L121

#5 lore raw
Quoted 8 lines · On Tue, 30 Jun 2026 at 14:33, Haase Moritz, DE-811 <[email protected]> wrote:

just saw that I originally disabled the build of the apps as part of the CMake 4 migration, since they were not packaged and upstream doesn't consider them ready to be shipped (see [0]). This hasn't changed, the same text is still there and there are no install() rules for them.

Should we still try to package them? Even if we did that, the test itself would likely also require patching, as it tries to execute the json_parse app using a relative path (out of the build dir I guess).

You can probably just add the required binary to the ptest package then into a location that mimics the build directory? Ptests are similarly copied manually out of ${B} so it would simply be treated as a test helper.

Alex